Comments on "the differentiation between grid spacing and resolution and their application to numerical modeling"

Abstract

The comments of Grasso are extended to show that the minimum wavelength that is resolved by a numerical model is typically longer than the shortest unfiltered wave in the model. This result occurs because the numerical grid is unable to accurately describe the amplitudes and horizontal derivatives of short wavelengths in the numerical solution. The numerical scheme may also introduce phase and dispersion errors that cause short wavelengths to be inaccurate.